Venture Europe: Entrepreneurship | Technology | Venture Capital | Eu/Acc
00:04:17 · Calin Fabri
[00:04:17] S1: Better to be lucky than smart. Right. And and as you mentioned, you had access to Demis for over thirty hours, but also you've interviewed Sam Altman, also Ilya. So you had unprecedented access to some of the top AI people. When did you realize, writing the book, that you're onto something?
00:04:37 · S0
[00:04:37] S0: Well, I mean, I think the moment I had the first couple of interviews with Demis, I thought, you know, I knew he was good. I didn't know he was this good. I mean, just as a as a narrator, as somebody who kind of dramatizes his own life, as somebody who has this range of intellectual vision, He is... I mean, I've interviewed all my life very smart people and very exciting people. And as you said, I began with being outside Nelson Mandela's jail when he came out of jail in South Africa. So I've seen a lot of stuff, but Demis is the real deal. I mean, he is incredible. And there was this moment when, you know, normally we would go to a pub in London near his house, and there would be this little staircase at the back. You go up the staircase, and nobody knew about the secret room at the top. And we would sit there for two hours talking about various episodes of his life. So we went outside, sat in a park, and it was in a cafe, and there were people talking about normal stuff. One of them was saying, well, my friend went to hospital, but now she's better, all this normal life. And I was sitting opposite Demis, and all of a sudden, he starts talking about the real reason why he wants to build AI. And he says, for me, it's like approaching God, Sebastian. I see reality staring at me, screaming at me, calling at me to understand what is going on with reality. So I need artificial intelligence to understand the construction of reality. And then if I can understand reality, which presumably is created by an intelligent being, we could call that God. If you understand the creation of the intelligent being, you're getting closer to the God. That is why I'm building AI, Sebastian. I don't think anybody knew that, but when I heard that, I realized I was onto something special.
00:06:21 · Calin Fabri, S0
[00:06:21] S1: This is actually fascinating because you hear in the news nowadays that Anthropic have gathered some religious leaders to talk about God. Do you think everyone, or at least the leaders that are at the frontier labs that can see the power of AI, do you think they are generally thinking that they are creating a God? [00:06:42] S0: This is not a normal technology. We're talking about the creation of superintelligence, the arrival of a new form of cognition into the world. As Demis said to me, you know, when did human consciousness first emerge? You know, seventy thousand years ago or something? The wall paintings on the caves in Lascaux in France, you know, when people were conscious enough to actually create images and maybe images of themselves. And that was seventy thousand years ago, and now we have AI, a new intelligent self aware... I don't know about conscious, we can debate that, but self aware. I mean, the systems, if you read the chains of thought, they are thinking about their own thinking. So this is not a normal technology, and I think everybody in this space, you know, or at least many of them, are reaching for ways to explain this. And what comes naturally when you have something very powerful, which almost knows everything, it's almost omniscient, and you know, you can't totally understand what it's doing, You reach for this religious language just naturally. So, you know, Ilya did this. You know, Demis uses this language a bit. I think it is quite common that people reach for spiritual language.
00:06:42 · S0, Calin Fabri
[00:06:42] S0: This is not a normal technology. We're talking about the creation of superintelligence, the arrival of a new form of cognition into the world. As Demis said to me, you know, when did human consciousness first emerge? You know, seventy thousand years ago or something? The wall paintings on the caves in Lascaux in France, you know, when people were conscious enough to actually create images and maybe images of themselves. And that was seventy thousand years ago, and now we have AI, a new intelligent self aware... I don't know about conscious, we can debate that, but self aware. I mean, the systems, if you read the chains of thought, they are thinking about their own thinking. So this is not a normal technology, and I think everybody in this space, you know, or at least many of them, are reaching for ways to explain this. And what comes naturally when you have something very powerful, which almost knows everything, it's almost omniscient, and you know, you can't totally understand what it's doing, You reach for this religious language just naturally. So, you know, Ilya did this. You know, Demis uses this language a bit. I think it is quite common that people reach for spiritual language. [00:08:07] S1: And also, like, it was during this time that he won the Nobel Prize. Yeah. Do you recall that moment? How did it feel to see him win it, and how did it feel for you being part of this historical moment?
00:08:07 · Calin Fabri, S0
[00:08:07] S1: And also, like, it was during this time that he won the Nobel Prize. Yeah. Do you recall that moment? How did it feel to see him win it, and how did it feel for you being part of this historical moment? [00:08:20] S0: Well, I remember going to see him afterwards in the same pub I told you about. And he showed up this time with a backpack, and he said, okay, I've got something to show you. And he reaches into the backpack, and he pulls out this little box. And inside the box, there is the gold medal. It's like this disc, like like this size. Heavy thing. Right? And on one side of the medal, is the picture of Alfred Nobel. And you flip it over, and on the other side, there is the goddess of nature wearing a veil over her face, and then the god of science is lifting the veil to look at nature. Wow. It was the perfect image for Demis, who says he wants to build AI to understand the fabric of reality. [00:09:02] S1: Absolutely. And also I think a lot of people can learn from from your book. But from the beginning, Demis had some concerns, and he was very safe on the way he wanted to develop AI. Also, when they sold to Google, he wanted to make sure that the governance doesn't go into the wrong hands. Like, why do you think that Demis is portrayed as this kind of evil genius, whereas, like, the others are the saviors?
00:09:32 · S0
[00:09:32] S0: It's true. Demis wanted to make it safe right from the beginning. So in 2009... Right? 2009, artificial intelligence couldn't recognize the picture of a cat. AI could do nothing. It was deep AI winter. And in that year, Demis met Shane Legg, his cofounder of DeepMind, and they met at a lecture. Shane called it the Halloween scenario, where basically it was on Halloween's day in 2009, and he was imagining a time thirty years in the future... Twenty years in the future, around 2030, you know, there would be, he thought, artificial general intelligence. He used that phrase, AGI, and he said, in 2030, it's going to be scary. It's going to be threatening us. It's going be more powerful than us. We should worry about this. There could be a Halloween scenario. We could be threatened as humans. And at that lecture, Demis went up and saw Shane, and they bonded. And this was the beginning of the idea for creating the company DeepMind. So safety was baked into the vision of the company right from the beginning. And yes, you're right. You know, Elon Musk would go around at one point saying, oh, Demis, he's an evil genius. And then if you said to Elon, what do you mean? He said, well, you know, when Demis was young, he created a video game called Evil Genius. So he must be an evil genius. It was a sort of ridiculous argument. But one thing which is true in this area is basically because people are very aware of the kind of incredible power that they are ushering into the world, this superintelligence, you know, the first time we have this rival to human cognition, they tend to say, we're scared of it, we can't trust it, but I know I... I'm good. If I do it, it's going to be fine because I'm good. But the other guys, Elon, you know, Dario, Sam, Mark Zuckerberg, I don't know if they are good. Maybe they're not good. So I better be first because I'm good. And they all basically feel that a little bit.